There is a moment in the development of any technology when it works well enough. The demo runs. The client doesn’t complain. The metrics are green. You could ship it.
Erhan Ciris has learned to be afraid of that moment.
As the founder and CEO of 4D Sight, a virtual advertising insertion company whose clients include UFC and WWE, Ciris operates in a space where the standard of success is essentially invisible, where the product either belongs completely inside the broadcast frame or it doesn’t belong at all. That constraint, he says, gives “good enough” a particular seductiveness that most people outside of visual technology don’t anticipate.
“In a fast-moving broadcast, at normal viewing distance, on a standard consumer screen, a virtual ad insertion that is eighty percent right will pass,” Ciris explains. “Most viewers won’t consciously register the problem. The impression gets counted. The client sees the report. Everyone moves on. The seam is still there. You just decided not to talk about it.”
Erhan Ciris has consistently made the other choice, the uncomfortable one, the slower one, the one that costs more and requires explaining to people who thought the work was finished. The choice to not show the seam even when no one in the room would have noticed if he had.
The clearest example came during the development of 4D Sight’s rendering pipeline. The system handled the majority of broadcast environments well, controlled arena lighting, consistent camera positions, stable surfaces. The problem surfaced in edge conditions: late afternoon outdoor events where natural light was dropping and arena fill lights were compensating, creating a transition zone where two different light sources competed and the scene’s apparent color temperature shifted minute to minute.
In those conditions, the insertion was off. Not catastrophically, not in a way that would appear in any measurement report. But if you paused the frame and looked carefully, which Ciris does, constantly, out of a habit built over years, you could see it. The virtual object carried a slightly different tonal character than the real surface beneath it. The light was wrong.
The team was ready to move on. A client deadline was close. By any reasonable external standard, the product was shippable.
Ciris rebuilt the lighting model.
What made that decision possible, he argues, is having a standard that exists independently of what anyone else in the room is measuring. “Most quality metrics are downstream of human perception,” he says. “You build a test, you define what passing looks like, you measure against that. But if the test was designed by people who were satisfied with eighty percent, the test will tell you you’ve succeeded when you haven’t.”
The standard he applies instead is deliberately human and deliberately unquantifiable: would a viewer notice? Not a client. Not an engineer. A person watching the broadcast who has no idea what virtual advertising insertion is and isn’t thinking about it at all, would they feel, in some way they couldn’t articulate, that something was slightly off?
That standard is harder to measure and nearly impossible to automate. It requires the people building the system to have internalized what they’re actually trying to achieve, and to be honest with each other and themselves when they haven’t achieved it.
In an early-stage company, that kind of stubbornness can look like a liability. Timelines slip. Explanations get complicated. Clients who were satisfied yesterday are now being told the work isn’t finished. But Ciris has come to believe that this willingness to hold the line, to say the metric passed and the client is happy and we’re still not done, is precisely what separates technology that endures from technology that’s merely competitive for a moment.
“The seam is always findable,” he says. “The question is whether you’re looking.”
At 4D Sight, looking is institutional. The invisible standard isn’t Ciris’s personal habit projected onto a company; it’s embedded in how the team evaluates work, what gets signed off, and what gets sent back. “This will pass” is not considered a finished sentence.
It’s a slower way to build. It is, by Ciris’s own accounting, a more expensive one. He believes, without ambivalence, that it’s the right one.
